Default effect
Claim assessed: A pre-selected option is chosen more often than if it had to be chosen actively.
Published · how we work
In short
The default is the option that applies if you do nothing: the pre-ticked box, the plan you’re enrolled in automatically, the factory setting. The default effect means people keep it far more often than they would choose it if they had to tick it themselves. Defaults are one of the most used tools of choice architecture, that is, changing how options are presented without ruling out the others, also called nudges. On average the effect is large, but it varies a lot from one situation to another.
What it says
The classic examples. Johnson and Goldstein (2003) compared countries by how organ donors are registered. Where you had to sign up actively, the share registered was in the tens of percent; where you were enrolled automatically and could opt out, it was above 90%. Madrian and Shea (2001) showed that employees were 50% more likely to join the retirement savings plan when enrollment was automatic than when they had to sign up themselves. We have not read these studies; we describe them from the meta-analysis below.
The first meta-analysis. Jachimowicz, Duncan, Weber and Johnson (2019) collected 58 datasets, with 73,675 participants in total, that directly compared a pre-selected option with one that had to be selected actively. The average effect was large: in studies with yes-or-no choices, the default was chosen on average 27 percentage points more often. In 46 cases the effect was positive, in 10 it was not significant, and in 2 the default was actually chosen less often. Their analysis suggests that if studies are missing from the literature, they are more likely the ones with large effects, but the authors urge caution in interpreting this. Defaults worked better for consumer purchases and worse for environmental decisions. They also worked better when they looked like a recommendation from whoever designed the choice, or like the normal state of things.
The second meta-analysis and the dispute. Mertens et al. (2022) analysed 447 results from over 200 publications on nudges, with over 2 million participants. Of the nine types of intervention, defaults had the largest effect. The authors did find signs, though, that successful studies get published more often than unsuccessful ones. Maier et al. (2022) reanalysed the same data with a method that corrects for this distortion. After the correction, they no longer found evidence that nudges work in general. For interventions that change the structure of the choice, the category defaults belong to, the evidence remained undecided: the effect dropped from 0.58 to 0.12. Using only the most precise result from each study, the estimated effect was 0.23, with weak evidence in its favour. In their reply, Mertens et al. agreed that selective publication is a real problem for the field.
Example
When you start a job, the form for the supplementary pension already has a 3% salary contribution ticked. Most employees leave it as it is, although few would have picked exactly 3% if they had to write in the amount themselves. The example is an illustration, not a study cited here.
How to apply it
The steps below are a practical approach we propose, based on the studies cited.
- Ask what you would choose if nothing were ticked. The default isn’t necessarily the best option for you.
- Look at who chose the default. Jachimowicz et al. found that defaults work better when they look like a recommendation. It’s worth asking whose interest the recommendation serves.
- Use the effect in your favour. An automatic transfer to savings, set up once, becomes a default you no longer have to choose each time.
- If you set rules for others, test them. The meta-analysis authors recommend testing each application, because the effect can be much smaller than the average, or even reversed.
Limits and nuances
- The variation is enormous. In Jachimowicz et al.’s meta-analysis, almost all of the difference between results comes from real differences between studies, not from chance. The average says little about how well a particular default will work.
- The two meta-analyses disagree about selective publication. Jachimowicz et al. found no sign that it inflates the effect of defaults; Maier et al. found strong distortion in the wider nudge literature. Maier et al. did not analyse defaults separately, so their verdict does not apply to them directly.
- The mechanisms are estimated, not measured. Two raters judged, from each study’s description, whether the default looked like a recommendation, the normal state or the easiest option. The authors warn that these judgements are approximate.
- What we read. We read Jachimowicz et al.’s (2019) meta-analysis, Maier et al.’s (2022) letter and Mertens et al.’s (2022) reply in full, all open access. From Mertens et al. (2022) we read the abstract, the results and the later published correction, whose figures we use. We have not read Johnson and Goldstein (2003) or Madrian and Shea (2001); we describe them from Jachimowicz et al.
Sources
- Brigitte C. Madrian, Dennis F. Shea (2001). The power of suggestion: Inertia in 401(k) participation and savings behavior
- Eric J. Johnson, Daniel Goldstein (2003). Do defaults save lives?
- Jon M. Jachimowicz, Shannon Duncan, Elke U. Weber, Eric J. Johnson (2019). When and why defaults influence decisions: a meta-analysis of default effects
- Stephanie Mertens, Mario Herberz, Ulf J. J. Hahnel, Tobias Brosch (2022). The effectiveness of nudging: A meta-analysis of choice architecture interventions across behavioral domains
- Maximilian Maier, František Bartoš, T. D. Stanley, David R. Shanks, Adam J. L. Harris, Eric-Jan Wagenmakers (2022). No evidence for nudging after adjusting for publication bias
- Stephanie Mertens, Mario Herberz, Ulf J. J. Hahnel, Tobias Brosch (2022). Reply to Maier et al., Szaszi et al., and Bakdash and Marusich: The present and future of choice architecture research
See also: Decoy effect, Endowment effect and status quo bias